HP3000-L Archives

February 1995, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Rodolfo Lopez <[log in to unmask]>
Reply To:
Rodolfo Lopez <[log in to unmask]>
Date:
Tue, 14 Feb 1995 15:50:00 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (64 lines)
Larry Boyd wrote:
>
>While technically IMAGE _is_ a network database, the question of joins does
>not actually reside on the type of database.  The question of joins has
more
>to do with the indexing features.  Denys is correct.  Indexing/keys is what
>makes joins work or not work well.  A poorly designed relational database
>(based on chosen indexes) will perform joins poorly, also.  So, the
question
>more relates to what types of indexes are necessary to execute properly
>performing joins.  Once this is defined, then the question is does the
>database you are using support these types of indexes?
>
>I have seen IMAGE, without b-trees, out perform "relational" databases
while
>doing joins.  With b-trees, performance can even be better.  I have heard
>all the arguments about how IMAGE is not a "relational" database,
therefore,
>it can't support the retrieval performance necessary in an SQL environment.
>I, respectfully, disagree.  In fact with basic b-trees, of any kind, in
>IMAGE, I, as well as others, can show that the performance in beyond
>expectation and beyond "relational" databases.  With one of the two
3rd-party
>indexing products, you additionally receive indexing features that are
*not*
>available in "relational" databases, and these features will increase
>performance even more.
>
 
Maybe Alfredo should help with some numbers. IMHO, no one has devoted so
much effort or knows stronger and better arguments and numbers to show why
Image is so good.
 
Even the throughput numbers of ALLBASE, which are better than MANY SQL
databases, fall short when compared to Image.
 
I heard from an inside contact at HP, that there is a performance comparison
between ALLBASE and other SQL databases on the 3000 for "INTERNAL PURPOSES
ONLY!".
 
The good part is that in most categories ALLBASE outshines the competition.
Why is the chart not public? Maybe because ALLBASE falls short in the
toolset around it as a development  platform and you need third party tools
to complement what others include as standard for a true development
scenario (you pay for what you get). Now the better part is that HP agrees
that a well designed ALLBASE application will only get around 80% of IMAGE's
performance.
 
I have not seen the paper myself, but I can share my experience. Recently we
had an on site class in Gupta's SQL Windows 5.0. As a test database, I took
an Image database with just a few datasets and created its DBE for client
access. The first day, I showed the instructor the setup for the class. We
did some inquiries on a dataset with several dozen thousand records. When he
noticed how fast the query was executed, he was completely AMAZED. He has a
lot of experience accessing AS/400s, DECs and PC based databases and knows
Sybase, Oracle, etc. He just exclaimed, "Wow, that was quick!".  Obviously,
as you probably are, I grinned.
 
As Larry states, good indexing is VERY, VERY  important. Image and its add
on third party utilities are outstanding here.
 
Regards,
[log in to unmask]

ATOM RSS1 RSS2